This patch allows hashing of memoryviews, as discussed on python-dev.

Does memory_hash() reject non-contiguous memory?

_Py_HashBytes() might be reused by unicode_hash() for PyUnicode_1BYTE_KIND.

Except these two minor nits, the patch looks good to me.

> Does memory_hash() reject non-contiguous memory?

It should, since it checks the strides array.

> _Py_HashBytes() might be reused by unicode_hash() for PyUnicode_1BYTE_KIND.

I don't really see how unicode hashing should be related to bytes
hashing (even though apparently an (outdated?) test checks for that).

I'm not sure that the "register" storage class specifier is still relevant with modern compilers: I'm pretty sure gcc ignores it unless -O0, and I think I've read somewhere Microsoft's compiler ignores it too.

> I'm not sure that the "register" storage class specifier is still
> relevant with modern compilers: I'm pretty sure gcc ignores it unless
> -O0, and I think I've read somewhere Microsoft's compiler ignores it
> too.

I don't think so either. I was lazily moving the original code, but I
could remove those bits.

New changeset 215690b782f8 by Antoine Pitrou in branch 'default':
Issue #13411: memoryview objects are now hashable when the underlying object is hashable.
http://hg.python.org/cpython/rev/215690b782f8

Fixed, thanks.

Reopening as a reminder that it isn't fixed yet in http://hg.python.org/features/pep-3118 .

Done in: http://hg.python.org/features/pep-3118/rev/508d5e3c579c